Parallel Testing

How it works

When parts of your task are independent, the agent can run them simultaneously instead of one at a time. Each parallel task gets its own browser tab and works independently. When everything finishes, you get a combined report.

This happens automatically when the agent sees an opportunity, but you can also ask for it directly - for example, "test the login page on mobile and desktop at the same time."

When the agent runs tests in parallel

• The same check needs to run across multiple pages
• Testing responsive layout across different screen sizes (mobile, tablet, desktop)
• Independent sections of your app that don't affect each other
• Gathering information from several pages before starting a task

When things run sequentially

Parallel execution doesn't help when steps depend on each other. The agent runs things one at a time when:

• Each step depends on the result of the previous one
• State needs to carry forward (like a shopping cart total)
• The task is a single page with a single flow

Login and shared state

Parallel tasks can share your login session, so they don't need to log in separately. They can also start with a clean browser for testing logged-out experiences. The agent picks the right approach based on what you're testing.